MOVOU (16*0)(KS), B0
	MOVOU (16*1)(KS), T0
	AESENC T0, B0
	MOVOU (16*2)(KS), T0
	AESENC T0, B0
	MOVOU (16*3)(KS), T0
	AESENC T0, B0
	MOVOU (16*4)(KS), T0
	AESENC T0, B0
	MOVOU (16*5)(KS), T0
	AESENC T0, B0
	MOVOU (16*6)(KS), T0
	AESENC T0, B0
	MOVOU (16*7)(KS), T0
	AESENC T0, B0
	MOVOU (16*8)(KS), T0
	AESENC T0, B0
	MOVOU (16*9)(KS), T0
	AESENC T0, B0
	MOVOU (16*10)(KS), T0
	CMPQ NR, $12

	for s := 1; s < 32; s++ {
		t0.Square(t0)
	}
	z.Mul(z, t0)
	t0.Square(z)
	t0.Mul(x, t0)
	for s := 0; s < 64; s++ {
		t0.Square(t0)
	}
	z.Mul(z, t0)
	t0.Square(z)
	t0.Mul(x, t0)
	for s := 0; s < 129; s++ {
		t0.Square(t0)
	}
	z.Mul(z, t0)
	t0.Square(z)
	t0.Mul(x, t0)
	for s := 0; s < 259; s++ {
		t0.Square(t0)
	}
	z.Mul(z, t0)
	for s := 0; s < 2; s++ {
		z.Square(z)
	}
	z.Mul(x, z)

	VL   (0*16)(x_ptr), T0
	VPDI $0x4, T0, T0, T0
	VL   (1*16)(x_ptr), T1
	VPDI $0x4, T1, T1, T1

	// First round
	VPERM T1, T0, SEL1, RED2    // d1 d0 d1 d0
	VPERM ZER, RED2, SEL2, RED1 // 0  d1 d0  0
	VSQ   RED1, RED2, RED2      // Guaranteed not to underflow

	VSLDB $8, T1, T0, T0
	VSLDB $8, T2, T1, T1

	VACCQ  T0, RED1, CAR1
	VAQ    T0, RED1, T0
	VACCCQ T1, RED2, CAR1, CAR2

	}
	t1.Mul(t1, t2)
	for s := 0; s < 2; s++ {
		t1.Square(t1)
	}
	t0.Mul(t0, t1)
	t1.Square(t0)
	for s := 1; s < 3; s++ {
		t1.Square(t1)
	}
	z.Mul(z, t1)
	t1.Square(z)
	for s := 1; s < 14; s++ {
		t1.Square(t1)
	}
	t0.Mul(t0, t1)
	t1.Square(t0)
	for s := 1; s < 17; s++ {
		t1.Square(t1)
	}
	z.Mul(z, t1)
	t1.Square(z)
	for s := 1; s < 48; s++ {
		t1.Square(t1)
	}
	z.Mul(z, t1)

)

var v0 p0.T
var v1 p1.T

type I0 interface {
	M(p0.T)
}

type I1 interface {
	M(p1.T)
}

type t0 int

func (t0) M(p0.T) {}

type t1 float64

func (t1) M(p1.T) {}

var i0 I0 = t0(0) // ok
var i1 I1 = t1(0) // ok

var p0i p0.I = t0(0) // ok
var p1i p1.I = t1(0) // ok

func main() {
	// check that reflect paths are correct,

	MOV	A1, _rt0_riscv64_freebsd_lib_argv<>(SB)

	// Synchronous initialization.
	MOV	$runtime·libpreinit(SB), T0
	JALR	RA, T0

	// Create a new thread to do the runtime initialization and return.
	MOV	_cgo_sys_thread_create(SB), T0
	BEQZ	T0, nocgo
	MOV	$_rt0_riscv64_freebsd_lib_go(SB), A0
	MOV	$0, A1
	JALR	RA, T0
	JMP	restore

nocgo:
	MOV	$0x800000, A0                     // stacksize = 8192KB

	//
	p256Sqr(t0, x, 1)
	p256Mul(t0, x, t0)
	p256Sqr(t1, t0, 2)
	p256Mul(t0, t0, t1)
	p256Sqr(t1, t0, 4)
	p256Mul(t0, t0, t1)
	p256Sqr(t1, t0, 8)
	p256Mul(t0, t0, t1)
	p256Sqr(t1, t0, 16)
	p256Mul(t0, t0, t1)
	p256Sqr(t0, t0, 32)
	p256Mul(t0, x, t0)
	p256Sqr(t0, t0, 96)
	p256Mul(t0, x, t0)
	p256Sqr(t0, t0, 94)

	p256Sqr(t1, t0, 1)
	if p256Equal(t1, x) != 1 {

	VPMULL	B0.D1, B2.D1, T1.Q1
	VPMULL2	B0.D2, B2.D2, T0.Q1
	VPMULL	B1.D1, B3.D1, T2.Q1
	VEOR	T0.B16, T2.B16, T2.B16
	VEOR	T1.B16, T2.B16, T2.B16
	VEXT	$8, ZERO.B16, T2.B16, T3.B16
	VEXT	$8, T2.B16, ZERO.B16, T2.B16
	VEOR	T2.B16, T0.B16, T0.B16
	VEOR	T3.B16, T1.B16, T1.B16
	VPMULL	POLY.D1, T0.D1, T2.Q1
	VEXT	$8, T0.B16, T0.B16, T0.B16
	VEOR	T2.B16, T0.B16, T0.B16
	VPMULL	POLY.D1, T0.D1, T2.Q1
	VEXT	$8, T0.B16, T0.B16, T0.B16

func (v *Element) Pow22523(x *Element) *Element {
	var t0, t1, t2 Element

	t0.Square(x)             // x^2
	t1.Square(&t0)           // x^4
	t1.Square(&t1)           // x^8
	t1.Multiply(x, &t1)      // x^9
	t0.Multiply(&t0, &t1)    // x^11
	t0.Square(&t0)           // x^22
	t0.Multiply(&t1, &t0)    // x^31
	t1.Square(&t0)           // x^62
	for i := 1; i < 5; i++ { // x^992
		t1.Square(&t1)
	}

	MOV	X2, S2 // S2 = RSP, S2 is unchanged by C code
	MOV	g_m(g), S3 // S3 = m
	// Save the old values on stack for reentrant
	MOV	m_vdsoPC(S3), T0
	MOV	T0, 24(X2)
	MOV	m_vdsoSP(S3), T0
	MOV	T0, 32(X2)

	MOV	RA, m_vdsoPC(S3)
	MOV	$ret-8(FP), T0 // caller's SP
	MOV	T0, m_vdsoSP(S3)

	MOV	m_curg(S3), T1
	BNE	g, T1, noswitch

	MOV	m_g0(S3), T1
	MOV	(g_sched+gobuf_sp)(T1), X2

noswitch: